Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-23 Thread Graham Leggett
On 21 May 2010, at 6:04 PM, Derek Atkins wrote: Because a situation arises when both of you need to make writes. Which copy is the authoritative copy? Using svn alleviates this somewhat, but isn't ideal. The authoritative copy belongs to whomever has the write token. Otherwise known as

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-21 Thread Derek Atkins
Graham Leggett minf...@sharp.fm writes: Why not? Because a situation arises when both of you need to make writes. Which copy is the authoritative copy? Using svn alleviates this somewhat, but isn't ideal. The authoritative copy belongs to whomever has the write token. It is really easy to

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Derek Atkins
z33...@gmail.com writes: I think I understand at least some of the reasons for GC being single-user, but I'm curious if there's been any more thought lately to building multi-user functionality into GnuCash in future? I believe there is an unstated goal of getting minimal multi-user support

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Derek Atkins
Graham Leggett minf...@sharp.fm writes: On 19 May 2010, at 11:47 PM, Per Kjeldaas wrote: Your solution to block the whole database is good enough for me. It's a real shame that a system fundamentally designed to offer multi user access to data should be crippled in such a fashion. In the

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Derek Atkins
Valdis Vītoliņš valdis.vitol...@odo.lv writes: I vote for not over-engineered solution and, I suppose simple lock for all database (only one can write) is OK. My clients/partners will not use SQL backend. They use Gnucash because it can be installed and used as many other standalone

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Graham Leggett
On 20 May 2010, at 6:32 PM, Derek Atkins wrote: Your solution to block the whole database is good enough for me. It's a real shame that a system fundamentally designed to offer multi user access to data should be crippled in such a fashion. In the process, virtually all reasons to use a SQL

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Derek Atkins
Graham Leggett minf...@sharp.fm writes: On 20 May 2010, at 6:32 PM, Derek Atkins wrote: Your solution to block the whole database is good enough for me. It's a real shame that a system fundamentally designed to offer multi user access to data should be crippled in such a fashion. In the

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-20 Thread Graham Leggett
On 20 May 2010, at 6:55 PM, Derek Atkins wrote: Well, over time one would hope that we can slowly rearchitect gnucash to be more aware of multi-user situations. I'm keen to do some of that work, as I need it directly. In any practical usage, even in it's simplest form, you start off small

We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Christian Stimming
Thanks, Herbert, for the link. The article raises one valid serious concern: The database backend does nothing to prevent multiple user access. This is bad because simultaneous access to the SQL database from multiple users will almost surely cause data loss. On the other hand, the

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread z33b0b
I think I understand at least some of the reasons for GC being single-user, but I'm curious if there's been any more thought lately to building multi-user functionality into GnuCash in future? Cheers, -- Erik Anderson On Wed, May 19, 2010 at 1:58 AM, Christian Stimming stimm...@tuhh.dewrote:

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Per Kjeldaas
I think there is a big difference between concurrent multi use and serial multi use. I would like to be able to use GnuCash with the same remote database, but from different computers at different times. This would be serial multi use. The system should ideally protect against any use that

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Tao Wang
On Wed, May 19, 2010 at 6:58 PM, Christian Stimming stimm...@tuhh.de wrote: The article raises one valid serious concern: The database backend does nothing to prevent multiple user access. This is bad because simultaneous access to the SQL database from multiple users will almost surely cause

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Valdis Vītoliņš
I vote for not over-engineered solution and, I suppose simple lock for all database (only one can write) is OK. My clients/partners will not use SQL backend. They use Gnucash because it can be installed and used as many other standalone applications. It is important, that Gnucash save data in

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Per Kjeldaas
Tao, Your solution to block the whole database is good enough for me. Thanks, Per. On May 19, 2010, at 2:26 PM, Tao Wang wrote: On Wed, May 19, 2010 at 6:58 PM, Christian Stimming stimm...@tuhh.de wrote: The article raises one valid serious concern: The database backend does nothing to

Re: We need to prevent multi-user access to the SQL backend (Re: New GnuCash article on LWN)

2010-05-19 Thread Graham Leggett
On 19 May 2010, at 11:47 PM, Per Kjeldaas wrote: Your solution to block the whole database is good enough for me. It's a real shame that a system fundamentally designed to offer multi user access to data should be crippled in such a fashion. In the process, virtually all reasons to use a